CN104778181B - A kind of method and its equipment measuring spectrum and library Spectral matching - Google Patents
A kind of method and its equipment measuring spectrum and library Spectral matching Download PDFInfo
- Publication number
- CN104778181B CN104778181B CN201410016003.4A CN201410016003A CN104778181B CN 104778181 B CN104778181 B CN 104778181B CN 201410016003 A CN201410016003 A CN 201410016003A CN 104778181 B CN104778181 B CN 104778181B
- Authority
- CN
- China
- Prior art keywords
- spectrum
- library
- parameter
- range
- measure
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01B—MEASURING LENGTH, THICKNESS OR SIMILAR LINEAR DIMENSIONS; MEASURING ANGLES; MEASURING AREAS; MEASURING IRREGULARITIES OF SURFACES OR CONTOURS
- G01B11/00—Measuring arrangements characterised by the use of optical techniques
- G01B11/02—Measuring arrangements characterised by the use of optical techniques for measuring length, width or thickness
-
- G—PHYSICS
- G03—PHOTOGRAPHY; CINEMATOGRAPHY; ANALOGOUS TECHNIQUES USING WAVES OTHER THAN OPTICAL WAVES; ELECTROGRAPHY; HOLOGRAPHY
- G03F—PHOTOMECHANICAL PRODUCTION OF TEXTURED OR PATTERNED SURFACES, e.g. FOR PRINTING, FOR PROCESSING OF SEMICONDUCTOR DEVICES; MATERIALS THEREFOR; ORIGINALS THEREFOR; APPARATUS SPECIALLY ADAPTED THEREFOR
- G03F7/00—Photomechanical, e.g. photolithographic, production of textured or patterned surfaces, e.g. printing surfaces; Materials therefor, e.g. comprising photoresists; Apparatus specially adapted therefor
- G03F7/70—Microphotolithographic exposure; Apparatus therefor
- G03F7/70483—Information management; Active and passive control; Testing; Wafer monitoring, e.g. pattern monitoring
- G03F7/70605—Workpiece metrology
- G03F7/70616—Monitoring the printed patterns
- G03F7/70625—Dimensions, e.g. line width, critical dimension [CD], profile, sidewall angle or edge roughness
-
- H—ELECTRICITY
- H01—ELECTRIC ELEMENTS
- H01L—SEMICONDUCTOR DEVICES NOT COVERED BY CLASS H10
- H01L22/00—Testing or measuring during manufacture or treatment; Reliability measurements, i.e. testing of parts without further processing to modify the parts as such; Structural arrangements therefor
- H01L22/10—Measuring as part of the manufacturing process
- H01L22/12—Measuring as part of the manufacturing process for structural parameters, e.g. thickness, line width, refractive index, temperature, warp, bond strength, defects, optical inspection, electrical measurement of structural dimensions, metallurgic measurement of diffusions
-
- H—ELECTRICITY
- H01—ELECTRIC ELEMENTS
- H01L—SEMICONDUCTOR DEVICES NOT COVERED BY CLASS H10
- H01L22/00—Testing or measuring during manufacture or treatment; Reliability measurements, i.e. testing of parts without further processing to modify the parts as such; Structural arrangements therefor
- H01L22/20—Sequence of activities consisting of a plurality of measurements, corrections, marking or sorting steps
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01B—MEASURING LENGTH, THICKNESS OR SIMILAR LINEAR DIMENSIONS; MEASURING ANGLES; MEASURING AREAS; MEASURING IRREGULARITIES OF SURFACES OR CONTOURS
- G01B2210/00—Aspects not specifically covered by any group under G01B, e.g. of wheel alignment, caliper-like sensors
- G01B2210/56—Measuring geometric parameters of semiconductor structures, e.g. profile, critical dimensions or trench depth
Landscapes
- Engineering & Computer Science (AREA)
- Manufacturing & Machinery (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Computer Hardware Design (AREA)
- Microelectronics & Electronic Packaging (AREA)
- Power Engineering (AREA)
- Investigating Or Analysing Materials By Optical Means (AREA)
- Testing Or Measuring Of Semiconductors Or The Like (AREA)
- Length Measuring Devices By Optical Means (AREA)
Abstract
The invention discloses a kind of methods measuring spectrum and library Spectral matching in optical critical dimension measuring apparatus, wherein：A. sample topography model M is established according to known sample message；B. it obtains each measure spectrum Sn of sample and finds out average measurement spectrum S0；C. it is farthest measure spectrum Sm to find out the measure spectrum corresponding to the mean square deviation MSE, maximum mean square deviation MSEm between each measure spectrum Sn and the average measurement spectrum S0；D. the range RS that matched spectrum is participated in library of spectra is determined according to the average measurement spectrum S0 and farthest measure spectrum Sm；E. each measure spectrum Sn is matched in the matching range RS, and obtains the match parameter Pn of each measure spectrum Sn.
Description
Technical field
The invention mainly relates to field of semiconductor manufacture.Particularly, it is related to a kind of in the optics pass of semiconductor fabrication process
In key size (Optical Critical-Dimension, OCD) measuring apparatus " semi-conductor silicon chip microtexture measurement spectrum "
Matched method and its equipment are carried out with " gross data library spectrum ".
Background technology
In 45nm in the development of lower node manufacturing technology, chip foundries (foundry) and integrated device manufacturer
(IDM) a large amount of measure is faced to challenge.Complete circuit function and guarantor only could be obtained by stringent size distribution control
Holder part high speed operation, therefore, successfully on-line measurement is to promoting yield and keeping profit extremely necessary.However, due to figure
The diminution of size and design rule, and continually introduce unique measurement request because of new processing procedure and new material so that it is online to survey
Amount is faced with completely new challenge.For realizing the quickly accurate demand for measuring fine structure, new imaging in new processing procedure new technology
Technology is constantly applied in the measurement of semiconductor technology pattern, such as scanning electron microscope (CD-SEM), atomic force microscope
(AFM) etc., realize the measurement to high-precision critical size (CD), gash depth size.But its measurement process complexity consumes
When, to sample may damaging property, cannot achieve on-line measurement.Optical thin film measuring instrument can be to the thin of multiple layers of different materials
Film thickness measures, but it cannot graphic structure region carries out pattern and lateral dimension measures to having.
Many size characteristics in semiconductor technology are reflected in specific tested region, which includes new technology and brand-new
The fine structure accurately controlled is needed in journey.OCD equipment is exactly based on the scattering letter for obtaining specific tested region periodic structure
Number and structure model to estimate the design parameter of structure.Critical size (CD) and other shapes may be implemented in OCD methods
The measurement of looks size.In specifically measuring case, many process obtained may need scanning electron microscopy simultaneously
Mirror, atomic force microscope, optical thin film measuring instrument etc. are respectively completed.Since OCD measurement methods have untouchable, non-demolition
It property while measuring multiple technology characteristics and can realize many advantages such as the on-line measurement of technique, therefore it is more and more extensive
Ground is applied in semi-conductor industry, and is rapidly developed towards the direction for being cured fine structure is more rapidly more accurately measured.
The principle that OCD is measured generally can be described as：Theoretical spectral corresponding with the pattern model of sample library is established, and
The best match that specific theoretical spectral realizes the measure spectrum obtained with OCD measuring apparatus is found from this library, so that it is determined that
Its structural parameters.Measure spectrum is the scattered light signal for the sample cycle medium that OCD measuring apparatus obtains, although passing through measurement
Spectrum cannot the directly anti-distribution for releasing medium, but the distribution of scattering medium can be established model and parameterized, and used
The method of numerical computations calculates the theoretical spectral library that the model corresponds to different parameters value, i.e., obtains and survey to OCD measuring apparatus
Light scattering when measuring spectrum carries out simulation calculation.The theoretical spectral with measure spectrum best match is searched out from theoretical spectral library
Corresponding special parameter, to determine the distribution of the medium of sample indirectly.
Spectral matching process establishes the pattern model v of sample cycle structure, pattern according to the technique information of sample first
Model is determined by structural parameters P (sample profile structural parameters).According to possibility offset of the sample to be tested in relation to technique, mould is set
The domain of walker for each structural parameters that type is included.We mark the number of each pattern model using call number index,
One vindexA corresponding specific pattern.It is assumed that vindexThere is I structural parameters P, each structural parameters are by J different numbers
Value.So i-th of structural parameters can be expressed as Pi, it has Ji different float values.Pass through the computing module of theoretical spectral
It can calculate by vindexThe theoretical spectral LS (v, λ) of its scattering of the pattern sample of decision.Finally by whole spectrum by index
Sequence composition library of spectra, a spectrum ls (v in library of spectraindex, λ) and correspond to a structure and morphology vindex.This modal data library
The total number of theoretical spectral LS (v, λ) isThat is the product of Ji to JI.The usual packet of theoretical spectral computing module
Include stringent wave coupling analysis theory Rigorus Coupled Waveguide Analysis (RCWA) algorithm.
A large amount of measure spectrum s (v, λ) is obtained by OCD measuring apparatus.Measurement noise is not considered, when we are in spectrum number
According to finding ls (v in libraryindex, λ) and=s (λ), then vindexIt is just the pattern for measuring sample, corresponding structural parameters P
It is the measurement parameter to be obtained.
Matching standard is usually using grade of fit GOF (Goodness of Fit) and mean square deviation MSE (Mean Square
Error) etc..Here we use the matching standard of MSE, and MSE values are smaller, show the more similar of two spectrum, if MSE=0,
Then show that this two spectrum are identical.MSE is defined as follows：
The wave-length coverage of theoretical spectral and measure spectrum is λ in above formula1To λNN number of discrete wavelength points composition.
With the needs of technique productions, measurement accuracy requirement is higher and higher, and the range of measurement is also increasing, each parameter
The centrifugal pump J of calculatingi(i=1 ..., I) it is just more.Simultaneously as technological requirement needs measurement structure pattern further fine, mould
The parameter that type needs is also more, i.e., I becomes larger.So that spectrum sum will be a very huge astronomical figure.
According to the above, traditional simple operations flow for building storehouse matching scheme (abbreviation traditional scheme) is as shown in Figure 1.
In step A01：To establish correlation model according to sample message；In step A02：Pattern is arranged according to known sample information to join
Several ranges of variables, the range of the variable is larger under normal circumstances, step-length is smaller；In step A03：It is established according to range of variables
Theoretical spectral database；In step A04：Board measurement obtains measure spectrum queue；In step A05：By measure spectrum team
Measure spectrum in row traverses entire theoretical spectral database one by one, finds best match spectra, to obtain measuring with each item
The optimal structural parameters value of Spectral matching；In step A06, judge whether to have matched all spectrum, if not matched
Continue to match next spectrum；The structural parameters value for matching and obtaining is confirmed in step A07 whether on the boundary of range of variables,
It is not the range of variables that is arranged every time when step A05 is matched all in appropriate range, if range of variables is too small to lead to shape
Looks parameter is matched to range of variables boundary, then the matching result obtained at this time is insincere, and return to step A02 is needed to reset variable
Range, and carry out building storehouse matching again to obtain optimal structural parameters value according to above-mentioned flow；In step A08, if matching
Complete all spectrum and structural parameters are all not matched on boundary, then match completion and export result.
It counts less model to relatively simple for structure, structure less or wavelength of counting, i.e.,：The less feelings of spectrum need to be calculated
Condition, calculated using local computer monokaryon using traditional scheme or multicomputer multinuclear at the same calculate can fast implement build library,
The calculating such as storehouse matching.But to the spectra database that model is more complicated or calculating spectrum is more, when storehouse matching need to occupy a large amount of
Between.Such as certain 2D naive model has 5 variables (i.e.：Structural parameters), each variable (structural parameters) has 11 centrifugal pump (double changes
The two-dimensional representation of amount such as Fig. 2-a, each lattice point represent one group of centrifugal pump), then sharing 11 in theoretical spectral library5=161051
Spectrum.Selecting the library file (161051 calculating spectrum) to carry out storehouse matching to its corresponding measure spectrum (such as 100 groups) needs
It will be for a long time.Also, library and measurement file are not analyzed before matching, just check whether that pattern is joined after matching
Number is matched to boundary, if being matched to boundary, matches needs and restarts, wastes a large amount of time.It is arranged in step A02
The range of structure variable is big, step-length is small, the spectral information of the centrifugal pump within the scope of all structure variables is included in database, still
In measure spectrum matching process in step A05, not all theoretical spectral information all necessarily participates in matching, it is assumed that
Know matched structural parameters value, then a certain range of theoretical spectral is matching effective range near the parameter value.According to this spy
Point, the present invention, using the information of measure spectrum, can improve storehouse matching by a relatively large margin while ensureing that computational accuracy is unaffected
Speed and validity.
Invention content
In order to solve the above-mentioned technical problem, the invention discloses one kind measures in optical critical dimension measuring apparatus
The method and the measuring apparatus of spectrum and library Spectral matching.
Disclose according to an aspect of the present invention it is a kind of measured in optical critical dimension measuring apparatus spectrum with
The method of library Spectral matching, wherein：A. sample topography model M is established according to known sample message；B. sample is obtained
Each measure spectrum Sn simultaneously finds out average measurement spectrum S0；C. each measure spectrum Sn and the average measurement spectrum are found out
The measure spectrum corresponding to mean square deviation MSE, maximum mean square deviation MSEm between S0 is farthest measure spectrum Sm；D. according to described flat
Equal measure spectrum S0 and farthest measure spectrum Sm determines the range RS that matched spectrum is participated in library of spectra；E. in the matching model
Matching each measure spectrum Sn in RS is enclosed, and obtains the match parameter Pn of each measure spectrum Sn.
Particularly, in the step D, further comprise：D1. the range of variables of the parameter of the pattern model M is set
Rp；D2. spectra database L is established；D3. from the spectra database L, the center for matching the average measurement spectrum S0 is obtained
Library spectrum LS0 and Center Parameter P0, and match the farthest library spectrum LSm and farthest parameter Pm of the farthest measure spectrum；d4.
Check whether the Center Parameter P0 and the farthest parameter Pm are matched on the boundary of the range of variables Rp：D41. work as matching
When on to the boundary of the range of variables Rp, then return to step d1 expands the range of variables Rp, and implements the step d2；
D42. when not being matched on the boundary of the range of variables Rp, then implementation steps d5；D5. the consolidated storage spectrum LS0 and
Mean square deviation MSELm between the farthest library spectrum LSm and average measurement spectrum S0 and farthest measure spectrum Sm it
Between mean square deviation MSEm in higher value be set as maximum mean square deviation MMSE (Maximum-Mean-Square-Error)；D6. root
The matching range RS is determined according to the Center Parameter P0, the consolidated storage spectrum LS0 and the maximum mean square deviation MMSE.
Particularly, in the step d6, further comprise：The model centered on the model corresponding to Center Parameter P0,
Calculate the mean square deviation MSEi of the library spectrum LSi and the consolidated storage spectrum LS0 of its ambient parameters Pi：As MSEi≤r*MMSE,
The weight of the library spectrum LSi is set as 1；As MSEi ＞ r*MMSE, the weight of the library spectrum LSi is set as 0；Wherein, the r
For regulation coefficient；The library spectrum LSi that weight is set as 1 is the library spectrum in the matching range RS；The parameter Pi is located at described
In range of variables Rp.
Particularly, further include step F：When the match parameter Pn obtained in the step E is still matched to the change
When measuring the borderline situation of range Rp, then return to step d1 expands the range of variables Rp, and implements the step d2；When
When the match parameter Pn obtained in the step E is not matched to the borderline situation of the range of variables Rp, output
The match parameter Pn.
Particularly, in the step d6, the r is the number that value range is 1-3.
Particularly, in the step d6, further include：When the simply connected for surrounding the Center Parameter value P0 in the model space
When the weight of the library spectrum LSi of any one of parameter Pi on high-dimensional island geometry surface is all 0, then the list
The weight for being connected to the library spectrum LSi of the Pi outside high-dimensional island geometry surface is all set as 0.
Particularly, in the step d6, the model centered on the model corresponding to Center Parameter P0, according to it is described in
The distance of the distance of heart parameter P0, by closely to the library spectrum LSi and the consolidated storage for far gradually calculating its ambient parameters Pi outward
The mean square deviation MSEi of spectrum LS0.
A kind of measuring apparatus of optical critical dimension is disclosed according to another aspect of the present invention, it is characterised in that the right to use
Profit requires any method in 1 to 7 to match measure spectrum with library spectrum, and exports the match parameter after matching.
Due to having carried out the analysis of weight distribution to the library spectrum in database in the present invention so that without to library when matching
In all library spectrum traversed, in the case where not reducing matching precision, greatly reduced the library used in actual match
The quantity of spectrum so that match time greatly declines.Whether the matching of library spectrum and measure spectrum is matched in the present invention
Boundary has carried out Primary Assay, has chosen and most matches most representative two spectrum and matched for the first time, greatly reduces
Measure spectrum is matched to the probability on the boundary of library spectrum.
Description of the drawings
By the way that hereafter the embodiment in conjunction with shown by attached drawing is described in detail, above-mentioned and other features of the invention
It will be apparent from, same or analogous label indicates same or similar step in attached drawing of the present invention；
Fig. 1 shows the method flow diagram of a traditional measure spectrum and library Spectral matching；
Fig. 2-a show that the two-dimensional representation of bivariate, Fig. 2-b are shown in the two-dimensional representation of bivariate, average
Measure spectrum S0 and conventional libraries matching result with the maximum spectrum Sm of average measurement spectrum S0 differences；
Fig. 3 is shown according to a kind of method flow diagram measuring spectrum and library Spectral matching disclosed in this invention；
Fig. 4 shows the process schematic for the library spectrum by taking the library spectrum of double structural parameters as an example being arranged weight；
Fig. 5 shows the pattern model and structural parameters of the sample in embodiment of the present invention；And
Fig. 6-a show that the distribution schematic diagram of measure spectrum, Fig. 6-b show the distribution schematic diagram of matching structural parameters.
Specific implementation mode
It, will be with reference to the appended attached drawing for constituting a present invention part in the specific descriptions of following preferred embodiment.Institute
Attached attached drawing, which has been illustrated by way of example, can realize specific embodiment.Exemplary embodiment is not intended to
Limit all embodiments according to the present invention.It should be noted that although described in the present invention with particular order has herein
The step of pass method, but this does not require that or implies and must execute these operations or necessary according to the particular order
Executing operation shown in whole could realize desired as a result, on the contrary, step described herein, which can change, executes sequence.
Additionally or alternatively, it is convenient to omit multiple steps are merged into step and executed by certain steps, and/or by a step
It is decomposed into execution of multiple steps.
For convenience of description, hereinafter own " parameter " all to refer to " structural parameters ".
Optical critical dimension measures analysis software (Optical Critical-Dimension Analysis at present
Software the Modeling and Design to semi-conductor silicon chip microstructure) can be achieved, build the operations such as library calculating, storehouse matching.Wherein, pass through
It carries out building library using RCWA and theoretical spectral database i.e. library file is calculated, a large amount of measure spectrum measured with practical board
Data are matched to obtain measuring sample parameters value.
Although at present usually used traditional matching scheme can obtain with measure spectrum optimal value of the parameter the most matched,
But and measure spectrum quantity larger in spectra database it is more when, all measure spectrums both participate in matching and every time matching all
All spectrum in library are needed to be traversed for, need largely to calculate the time.However, substantially participating in matched measure spectrum and being often
The measure spectrum of a kind of sample, the difference between measure spectrum are very small.The optimal value of the parameter difference of all measure spectrums nor
Often small, matching only needs to find several spectrum near the parameter and participates in calculating.Matching only needs to use theory in this way
Partial spectrum in library of spectra greatly reduces and participates in matched theoretical spectral number, reduces and calculates the time, and the present invention carries accordingly
For following scheme to implement the matching of measure spectrum and library spectrum.
Fig. 3 shows according to a kind of method flow diagram measuring spectrum and library Spectral matching disclosed in this invention,
It includes mainly：The more accurate spectra database with larger range of variables is established according to sample message；To required
The measure spectrum matched is analyzed；The range for participating in matched library spectrum is determined according to the analysis result of measure spectrum；It is matching
Measure spectrum progress is matched one by one in range and etc..Concrete operations flow is as follows：
Step B01~B03：Model and spectra database are established according to sample message.
It is similar with traditional scheme in step B01, correlation model M is established according to sample message；In step B02, according to
Known sample information, is arranged the range of variables Rp of the parameter in the pattern model M, and range of variables Rp and traditional scheme are set
Set that mode is identical, step-length is also identical with traditional scheme；In step B03, established according to set range of variables Rp corresponding
Spectra database L, the spectrum in spectra database L is library spectrum.
In one embodiment disclosed by the invention, to certain 2 dimension naive model modeling, there are 5 parameters, when parameter
When range of variables is set as each parameter with 11 centrifugal pumps, then sharing 11 in the L of theoretical spectral library5=161051 spectrum.
The two-dimensional representation by taking two parameter CD (critical size), HT (height) as an example is shown in Fig. 2-a, wherein intersecting point coordinate is
Centrifugal pump, the intersection point spacing of variable indicate step-length.
In step B04~B05：Analyze measure spectrum information.
In step B04, obtain measure spectrum queue Sn (n=1,2 ..., x, x is of spectrum in measure spectrum queue
Number), and to carry out average operation to obtain average measurement spectrum S0.
In step B05, each measure spectrum Sn in measure spectrum queue and each between averaged spectrum S0 is found out
Mean square deviation MSE, the maximum measure spectrum of mean square deviation with averaged spectrum S0 are farthest measure spectrum Sm.The maximum mean square deviation is
MSEm.For example, at one in the measure spectrum queue with 3 measure spectrums, first between measure spectrum and averaged spectrum
MSE is 1, and second MSE between measure spectrum and averaged spectrum is 2, and the MSE between third measure spectrum and averaged spectrum is 3,
Then third measure spectrum is farthest measure spectrum, MSEm=3.
In step B06~B08：Center and the radius of matching range Rs are determined according to the measure spectrum information of analysis.
We indicate the distance between two spectrum with the MSE values between two spectrum, then one can be found out most
Big mean square deviation MMSE, all measure spectrum Sn are distributed in centered on S0, and radius is in the field of MMSE.Due at primary
Measure spectrum in matching is close sample, and therefore, the radius in this field is very small.
According to the principle of storehouse matching, it is assumed that measure and error is not present, when finding a library spectrum LSn in spectra database
When being exactly matched with measure spectrum Sn, i.e. when MSE (LSn, Sn)=0, the model v corresponding to LSnnMeasured by being fully equivalent to
Sample.And when measurement is there are when error, although vnIt is fully equivalent to measure sample, and when library spectrum LSn is matched with Sn, MSE
The value that (LSn, Sn) is also not exactly equal to 0, MSE (LSn, Sn) is measurement error.Therefore under normal circumstances, what storehouse matching obtained
MSE values are all close to 0, it is believed that its value is measurement error.
Difference between measure spectrum, which is mainly derived from, to be measured caused error and measures the difference between sample, and is counted
The difference of set model is then entirely derived from according to the difference between the theoretical spectral in library.MSE between two theoretical spectrals
The difference between two corresponding models is marked in value, and the MSE gaps of even two theoretical spectrals the big, indicates corresponding
Difference between two models is bigger.
Assuming that measure no error, the difference of two measure spectrums Sa and Sb between them is MSE (Sa, Sb), they are complete
Complete matched library spectrum is respectively LSa and LSb, i.e. MSE (LSa, Sa)=0, MSE (LSb, Sb)=0, library spectrum LSa and Sa are complete
Exactly the same, LSb is identical with Sb.Therefore, MSE (LSa, LSb)=MSE (Sa, Sb) measures sample and set structure mould
Type is identical.
When measurement is there are when error, and MSE (LSa, Sa) → 0 (MSE (LSa, Sa) close to 0), MSE (LSb, Sb) → 0
(MSE (LSb, Sb) close to 0), and library spectrum LSa and Sa, the difference between LSb and Sb is measurement error, under normal circumstances
Measurement error is small value, the difference then smaller, therefore MSE (LSa, LSb) ≈ MSE of the measurement error of 2 measurements of same system
(Sa,Sb).Therefore the distributed intelligence of measure spectrum is almost identical as the distributed intelligence of their library spectrum exactly matched.
By taking 2 dimension parameter CD and HT as an example, as shown in Fig. 6-a, the distance between S0 and Sm are MSEm, their best match
Parameter distribution be P0 (CD0, HT0) and Pm (CDm, HTm), mean square deviation between corresponding library spectrum LS0 and LSm is MSELm,
MSEm≈MSELm.The present invention is based on this principle to the analysis of library spectrum, when a large amount of measure spectrum is distributed in centered on S0,
When radius is in the field of MSEm, their the library spatial distribution exactly matched by S0 centered on exactly matching spectrum LS0,
Radius is in the field of MSELm or so.Therefore the library spectrum of the best match of corresponding a large amount of measure spectrums is in this field
Interior, our radiuses appropriate for being exaggerated this field in practical operation, matching only needs in this field can find
Best match spectra, without the theoretical spectral being still in matching library outside this field.
In step B06, conventionally found out in library of spectra L with the matched library spectrum of spectrum S0 and Sm, S0's
Centered on best match library spectrum centered on the result of the corresponding parameter of library spectrum LS0, LS0 parameter P0, Sm best match library
Spectrum is that the result of the corresponding parameter of farthest library spectrum LSm, LSm is farthest parameter Pm, and Fig. 2-b show that P0 and Pm is sat in two dimension
Position view in mark.
In step B07, whether inspection parameter P0 and Pm are matched on the boundary of range of variables Rp, if being matched to change
It measures and then illustrates that there may be more preferably match parameters other than range of variables Rp on the boundary of range Rp, current matching result is not
It is credible, it needs to reset range of variables Rp in return to step B02.Range of variables Rp needs are enlarged, to confirm variable model
Enclose whether the parameter outside Rp is more preferably match parameter.If parameter P0 and Pm are not matched on the boundary of range of variables Rp,
Then continue implementation steps B08.
In step B08, the mean square deviation MSELm of consolidated storage spectrum LS0 and farthest library spectrum LSm are calculated, calculates average survey
The mean square deviation MSEm between spectrum S0 and farthest measure spectrum Sm is measured, takes MSEm with the greater in MSELm as maximum square
Poor MMSE.Radiuses of the MMSE as the matching range Rs centered on consolidated storage spectrum LS0.Wherein, matching range Rs is
The range of matched spectrum is participated in library of spectra L.
In step B09：It is specific to determine matching range Rs.
In step B09, by defining the weight of library spectrum come specific each library spectrum set in spectra database L
Whether in matching range Rs.Parameter centered on Center Parameter P0 corresponding to library spectrum LS0.Corresponding to Center Parameter P0
Model centered on model gradually calculates the MSE between the library spectrum and LS0 of its ambient parameters Pi outward, as MSE≤r*MMSE,
Define this library spectral weight be 1, on the contrary it is then define this library spectral weight be 0.Wherein, r is regulation coefficient, to adjust
Size with range Rs.In a preferred embodiment, r is chosen for the numerical value of 1-3.Pi is arbitrary around model center P0
One parameter point (i=1,2 ..., y, y is the number of the parameter around P0).When surrounding the Center Parameter value in the model space
The weight of the library spectrum LSi of any one of parameter Pi on the high-dimensional island geometry surface of simply connected of P0 is all 0
When, calculating terminates, and the weight for defining the library spectrum LSi of the Pi of this surface external is all 0.Lattice point on this surface is library spectrum
MSE with the consolidated storage spectrum LS0 is the centrifugal pump on the outside of the contour of r*MMSE.By taking 2 dimension parameters as an example, in Fig. 4-4C
Soft dot is the lattice point on simply connected multidimensional face, and dotted-line ellipse is that the MSE in the case of 2 dimension parameters is contour equal to r*MMSE
Line.Under normal circumstances, the shape of contour is the geometry of the high-dimensional island of simply connected.
Fig. 4 shows the process signal for the library spectrum by taking the library spectrum of double structural parameters (CD and HT) as an example being arranged weight
Figure.In Figure 4 A, the stain of the model centered on the model corresponding to Center Parameter P0, center is P0.First choose the nearest of P0
Marked as 1,2,3,4 and with the lattice point that empty circles mark in adjoint point Pi, i.e. Fig. 4-4A, calculate the library spectrum of this 4 parameters with
MSE between LS0, and according to the weighted value (power of the four library spectrum in this example of this determining four library spectrum of above-mentioned rule
1) weight values are.Then secondary Neighbor Points for choosing P0, i.e. marked as 5,6,7,8 and with the lattice point of empty circles label in Fig. 4-4B,
And the weighted value of this four spectrum is recorded, and so on.As shown in figures 4-4 c, when surrounding the Center Parameter value in the model space
When the weight of the library spectrum LSi of any one of parameter Pi on the simply connected surface of P0 is all 0, calculating terminates, and defines
The weight of the library spectrum LSi of the Pi of this surface external is all 0.As shown in Fig. 4-4C, the solid dot of black indicates result of calculation
The parameter for the library spectrum that weight is 1, hollow dots indicate that result of calculation weight is the parameter of 0 library spectrum, remaining crosspoint table
Show the parameter for directly defining the library spectrum that weight is 0.Library spectrum in Fig. 4-4D corresponding to solid black point parameter is to need
Matched library spectrum is carried out with test spectral.
In step B10~B11：Matching is implemented to measure spectrum Sn in matching range Rs, and confirms the ginseng being matched to
Whether number Pn is on the boundary of range of variables Rp.
In step B10, the spectra database and the measure spectrum Sn in measure spectrum queue for having been added to weight are utilized
It is matched, the library spectrum that wherein weight is 0 is not involved in matching, and the library spectrum that weight is 1 participates in matching.In Fig. 4-4D, mark
The library spectrum corresponding to the parameter of solid black color dot is denoted as to be matched, and the library light corresponding to the parameter that remaining crosspoint represents
Spectrum is not involved in matching.
In step B11, all measure spectrum Sn have been matched in matching range Rs, matching result are checked, if still having
Parameter Pn is matched to the boundary of range of variables Rp, then return to step B02 changes range of variables, optimizes library of spectra L, and again
Match.If matching result is normal, storehouse matching is completed, and the matched parametric results of each obtained measure spectrum are each measurement sample
The corresponding size parameter values of product.
Below will by comparing using the method for the present invention equipment with using traditional scheme equipment matching process with
Illustrate the operating process and advantageous effect of the method for the present invention with speed.
By taking naive model shown in fig. 5 as an example, there are 3 structural parameters, i.e.,：The critical size (MCD) of sample, height
(HT), side wall angle (SWA).The range of variables Rp of setup parameter：MCD has 71 centrifugal pumps, HT to have 21 centrifugal pumps, SWA to have 13
A centrifugal pump.Therefore building library obtains spectra database L1.19383 spectrum are total up in library.It is a total of in measure spectrum queue
71 similar samples, each sample have 10 measure spectrums.Owned by directly traversing in L1 using the equipment of traditional scheme
Spectroscopic data matches 710 measure spectrums one by one, to obtaining the optimal structural parameters value of sample；
According to a kind of measuring apparatus of optical critical dimension disclosed in this invention according to above-mentioned side disclosed in this invention
Method carries out analysis and the averaged spectrum S0 that 710 spectrum is calculated to 710 measure spectrums；Then it acquires and averaged spectrum
The maximum farthest measure spectrum Sm and corresponding MSEm of S0 differences；Then by S0 and Sm conventional methods and spectra database L1 into
Row matching has obtained best matching library spectrum LS0 and LSm and optimum profile parameter P0 and Pm；Later, check that P0 and Pm is
On the boundary of the no range of variables Rp for being matched to parameter (it is not matched to boundary in present case, needs to repair if being matched to boundary
Change the range of variables of parameter, and re-start and build storehouse matching)；Then the MSELm values of LS0 and LSm are found out and compared with MSEm, are taken
Higher value is maximum mean square deviation MMSE；It is gradually extrapolated centered on P0, calculates the mean square deviation of the library spectrum and LS0 in the L1 of library
MSE chooses r=1.5 in this example, and as MSE≤1.5*MMSE, it is 1 to define the theoretical spectral weight of this, is otherwise 0；This case
Finally obtained 2580 theoretical spectrals in example is the library spectrum that weight is 1, and therefore, actual participation is matched in the present invention for this
Library spectrum number is 13.31% of library spectrum number in traditional scheme；Measure spectrum is matched with the library spectrum that weight is 1；Most
If being matched all measure spectrums complete output as a result, obtained match parameter Pn is not matched to boundary afterwards, this is completed
Secondary storehouse matching work, obtained optimum matching parameters Pn are the values of the structural parameters for measuring sample.
Table -1 lists the correction data of the match time of conventional method and the method for the present invention, there it can be seen that although
Invention increases to measure file and theoretical spectral database analysis, but its occupied time relative to storehouse matching when
Between it is considerably less.With increasing for number of files is measured, the time that the method matching based on the present invention can save is also more.
Table -1：The comparison of the match time of traditional matching process and the present invention
| Test spectral quantity | Traditional match time (second) | Match time (second) of the invention |
| 10 | 12.07 | 3.10 |
| 71 | 93.49 | 60.39 |
| 142 | 184.86 | 122.41 |
| 213 | 272.10 | 163.30 |
| 355 | 535.57 | 281.94 |
| 710 | 1251.25 | 629.46 |
It is obvious to a person skilled in the art that invention is not limited to the details of the above exemplary embodiments, Er Qie
In the case of without departing substantially from spirit or essential attributes of the invention, the present invention can be realized in other specific forms.Therefore, no matter
How from the point of view of, the present embodiments are to be considered as illustrative and not restrictive.In addition, it will be evident that one word of " comprising " not
Exclude other elements and step, and wording "one" be not excluded for plural number.The multiple element stated in device claim also may be used
To be realized by an element.The first, the second equal words are used to indicate names, and are not represented any particular order.
Claims (8)
1. a kind of method measuring spectrum and library Spectral matching in optical critical dimension measuring apparatus, wherein：
A. sample topography model M is established according to known sample message；
B. it obtains each measure spectrum Sn of sample and finds out average measurement spectrum S0；
C. the mean square deviation MSE between each measure spectrum Sn and the average measurement spectrum S0, maximum mean square deviation are found out
Measure spectrum corresponding to MSEm is farthest measure spectrum Sm；
D. the range that matched spectrum is participated in library of spectra is determined according to the average measurement spectrum S0 and farthest measure spectrum Sm
RS；
E. each measure spectrum Sn is matched in the matching range RS, and obtains the matching of each measure spectrum Sn
Parameter Pn.
2. according to the method described in claim 1, wherein, in the step D, further comprising：
D1. the range of variables Rp of the parameter of the pattern model M is set；
D2. spectra database L is established；
D3. from the spectra database L, the consolidated storage spectrum LS0 for matching the average measurement spectrum S0 and center ginseng are obtained
Number P0, and match the farthest library spectrum LSm and farthest parameter Pm of the farthest measure spectrum；
D4. check whether the Center Parameter P0 and the farthest parameter Pm are matched on the boundary of the range of variables Rp：
D41. when being matched on the boundary of the range of variables Rp, then return to step d1 expands the range of variables Rp, and
Implement the step d2；
D42. when not being matched on the boundary of the range of variables Rp, then implementation steps
d5；
D5. mean square deviation MSELm and the average measurement between the consolidated storage spectrum LS0 and the farthest library spectrum LSm
The higher value in mean square deviation MSEm between spectrum S0 and the farthest measure spectrum Sm is set as maximum mean square deviation MMSE；
D6. the matching model is determined according to the Center Parameter P0, the consolidated storage spectrum LS0 and the maximum mean square deviation MMSE
Enclose RS.
3. according to the method described in claim 2, wherein, in the step d6, further comprising：
The model centered on the model corresponding to Center Parameter P0 calculates the library spectrum LSi of its ambient parameters Pi and the center
The mean square deviation MSEi of library spectrum LS0：
As MSEi≤r*MMSE, the weight of the library spectrum LSi is set as 1；
As MSEi ＞ r*MMSE, the weight of the library spectrum LSi is set as 0；
Wherein, the r is regulation coefficient；The library spectrum LSi that weight is set as 1 is the library spectrum in the matching range RS；It is described
Parameter Pi is located in the range of variables Rp.
4. according to the method described in claim 3, further including step F wherein：
When the match parameter Pn obtained in the step E still has the borderline situation for being matched to the range of variables Rp
When, then return to step d1 expands the range of variables Rp, and implements the step d2；
When the match parameter Pn obtained in the step E is not matched to the borderline situation of the range of variables Rp
When, export the match parameter Pn.
5. according to the method described in claim 4, wherein, in the step d6, the r is the number that value range is 1-3.
6. according to the method described in claim 4, wherein, in the step d6, further including：
When any one on the high-dimensional island geometry surface of simply connected for surrounding the Center Parameter value P0 in the model space
When the weight of the library spectrum LSi of a parameter Pi is all 0, then outside the high-dimensional island geometry surface of the simply connected
The weight of the library spectrum LSi of Pi is all set as 0.
7. according to the method described in claim 4, wherein, in the step d6, being with the model corresponding to Center Parameter P0
Center model, according to the distance at a distance from the Center Parameter P0, by closely to far gradually calculating its ambient parameters Pi's outward
The mean square deviation MSEi of library spectrum LSi and the consolidated storage spectrum LS0.
8. a kind of measuring apparatus of optical critical dimension, it is characterised in that use any method pair in claim 1 to 7
Measure spectrum is matched with library spectrum, and exports the match parameter after matching.
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410016003.4A CN104778181B (en) | 2014-01-14 | 2014-01-14 | A kind of method and its equipment measuring spectrum and library Spectral matching |
| US14/594,368 US20150198434A1 (en) | 2014-01-14 | 2015-01-12 | Method and apparatus for measuring critical dimension of semiconductor |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410016003.4A CN104778181B (en) | 2014-01-14 | 2014-01-14 | A kind of method and its equipment measuring spectrum and library Spectral matching |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN104778181A CN104778181A (en) | 2015-07-15 |
| CN104778181B true CN104778181B (en) | 2018-08-10 |
Family
ID=53521104
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410016003.4A Active CN104778181B (en) | 2014-01-14 | 2014-01-14 | A kind of method and its equipment measuring spectrum and library Spectral matching |
Country Status (2)
| Country | Link |
|---|---|
| US (1) | US20150198434A1 (en) |
| CN (1) | CN104778181B (en) |
Families Citing this family (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117219530A (en) * | 2017-02-27 | 2023-12-12 | 诺威有限公司 | Method for predicting E-test structural performance |
| CN111637849B (en) * | 2020-05-29 | 2021-11-26 | 上海精测半导体技术有限公司 | Method, device and equipment for measuring morphology parameters |
| CN114861113A (en) * | 2021-01-20 | 2022-08-05 | 睿励科学仪器（上海）有限公司 | Method and device for acquiring sample parameter information |
| CN113091626B (en) * | 2021-03-29 | 2023-11-03 | 长鑫存储技术有限公司 | Method for measuring film thickness of semiconductor device |
| US11867497B2 (en) | 2021-03-29 | 2024-01-09 | Changxin Memory Technologies, Inc. | Method for measuring film thickness of semiconductor device |
| CN113295090A (en) * | 2021-04-08 | 2021-08-24 | 睿励科学仪器（上海）有限公司 | Method and device for comparing spectra |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101950361A (en) * | 2010-09-06 | 2011-01-19 | 中国科学院遥感应用研究所 | Adaptive extraction method of remote sensing image thematic information based on spectrum matching degree |
| CN103155110A (en) * | 2010-08-06 | 2013-06-12 | 应用材料公司 | Techniques for matching measured spectra to reference spectra for in-situ optical monitoring |
-
2014
- 2014-01-14 CN CN201410016003.4A patent/CN104778181B/en active Active
-
2015
- 2015-01-12 US US14/594,368 patent/US20150198434A1/en not_active Abandoned
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103155110A (en) * | 2010-08-06 | 2013-06-12 | 应用材料公司 | Techniques for matching measured spectra to reference spectra for in-situ optical monitoring |
| CN101950361A (en) * | 2010-09-06 | 2011-01-19 | 中国科学院遥感应用研究所 | Adaptive extraction method of remote sensing image thematic information based on spectrum matching degree |
Also Published As
| Publication number | Publication date |
|---|---|
| US20150198434A1 (en) | 2015-07-16 |
| CN104778181A (en) | 2015-07-15 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104778181B (en) | A kind of method and its equipment measuring spectrum and library Spectral matching | |
| TWI719804B (en) | Method of optical metrology, computer program product, metrology module, target design file, landscape and metrology measurements of targets | |
| JP6463701B2 (en) | Method and system for optimizing optical inspection of patterned structures | |
| US10726169B2 (en) | Target and process sensitivity analysis to requirements | |
| CN107533995B (en) | Hot spot monitoring based on model | |
| TWI649641B (en) | Method for determining correction of patterning process, component manufacturing method, control system for lithography apparatus, and lithography apparatus | |
| JP5855728B2 (en) | Method, apparatus and medium for designing a measurement target | |
| JP4302965B2 (en) | Semiconductor device manufacturing method and manufacturing system thereof | |
| KR101943593B1 (en) | Method and system for providing a quality metric for improved process control | |
| CN101542280B (en) | Transforming metrology data from a semiconductor treatment system using multivariate analysis | |
| US8793638B2 (en) | Method of optimizing design for manufacturing (DFM) | |
| KR20200050478A (en) | Metrology method, computer product and system | |
| CN105874387A (en) | Method and apparatus for design of a metrology target | |
| US20170102623A1 (en) | Methods and Apparatus for Simulating Interaction of Radiation with Structures, Metrology Methods and Apparatus, Device Manufacturing Method | |
| CN104736962A (en) | Method for estimating and correcting misregistration target inaccuracy | |
| US20110060552A1 (en) | Measuring Apparatus, Measuring Coordinate Setting Method and Measuring Coordinate Number Calculation Method | |
| TWI637243B (en) | Metrology method and apparatus, computer program and lithographic system | |
| JP2020534558A (en) | Patterning process control method, device manufacturing method | |
| CN114930117A (en) | Combining physical modeling with machine learning | |
| US11366397B2 (en) | Method and apparatus for simulation of lithography overlay | |
| TWI660403B (en) | Method and apparatus for pattern fidelity control | |
| CN106289095B (en) | Critical size measurement method and equipment based on preceding value | |
| CN104572742B (en) | It is a kind of to be used to improve the method and device that theoretical spectral database creates speed | |
| Gao et al. | Few-shot Bayesian performance modeling for silicon photonic devices under process variation | |
| Wang et al. | Neural network driven sensitivity analysis of diffraction-based overlay metrology performance to target defect features |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| EXSB | Decision made by sipo to initiate substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| PE01 | Entry into force of the registration of the contract for pledge of patent right | ||
| PE01 | Entry into force of the registration of the contract for pledge of patent right |
Denomination of invention: Method and equipment thereof for matching measurement spectrum and library spectrum Effective date of registration: 20190121 Granted publication date: 20180810 Pledgee: Shanghai Xingcheng Investment Management Co., Ltd. Pledgor: Raintree Scientific Instruments (Shanghai) Corporation Registration number: 2019310000002 |
|
| PC01 | Cancellation of the registration of the contract for pledge of patent right | ||
| PC01 | Cancellation of the registration of the contract for pledge of patent right |
Date of cancellation: 20191029 Granted publication date: 20180810 Pledgee: Shanghai Xingcheng Investment Management Co., Ltd. Pledgor: Ruili Scientific Instruments (Shanghai) Co., Ltd. Registration number: 2019310000002 |